Those who may be dismissive of the technique of “DeHalo_alpha()”, because it produces a softer (less sharp) image, should keep in mind that halo-ing was not a deliberate, stand-alone addition to the picture. Rather it was an artifact of a standard, sub-standard sharpening technique. Reversing the halo-ing itself has the side-effect of removing the sharpening that created it. The only need thereafter is to put the sharpening back in … using a better sharpener.
So, to the previous proof-of-concept using DeHalo_alpha(), I’ve added a mild application of “LimitedSharpen()” to bring the sharpness back to the original picture’s appearance without it’s accompanying halo-ing.
(NOTE: Both DeHalo_alpha() and LimitedSharpen() have bunches of vaguely-defined numbers to control their operations. Ultimately, one must experiment to determine how each number affects the final result, and in what way. I did much back-and-forth until finally settling on these numbers used for this proof-of-concept. They should not, necessarily, be considered definitive.)
[again, the view is x2 magnified for inspection]
DeHalo_alpha( rx=1.5, ry=2.85, darkstr=1.2, brightstr=1.2 )
LimitedSharpen( ss_x=3.0, ss_y=3.0, strength=150, Lmode=2, soft=true, edgemode=1, exborder=4 )
Tweak( cont=1.0, bright=-5, coring=false )
As can be seen, the “after” picture looks just like the “before” picture, sans the before’s pronounced halo-ing. Even more, a side benefit of reversing the sub-standard sharpening has reduced it’s pronounced jaggies. With the picture now properly sharpened, those jaggies have remained reduced. It looks like a win-win solution to me!